Neurocognitive outcomes in long-term survivors of Wilms tumor: A report from the St. Jude Lifetime Cohort

Abstract

Purpose:

To examine prevalence and predictors of neurocognitive outcomes, social attainment, emotional distress and health-related quality of life (HRQOL) in long-term survivors of pediatric Wilms tumor (WT).

Methods:

158 WT-survivors (59% female; mean[SD] age 33[9.1] years; time since diagnosis 29[9.1] years) and 354 community controls (55.6% female; 35[10.2] years) completed comprehensive neuropsychological testing and physical examination, including echocardiography/electrocardiography, pulmonary function tests and endocrine evaluation. Self-report of emotional distress, HRQOL and social attainment were collected. Impairment was defined in relation to both controls and normative data. Generalized linear models were developed to examine impact of treatment and chronic health conditions on outcomes.

Results:

WT-survivors performed poorer than norms and controls in 6 of 16 cognitive variables and 1 of 8 HRQOL variables, with scores ranging from −0.64 (mathematics) to −0.21 (verbal fluency) standard deviations below expectations. Compared to controls, WT-survivors were less likely to graduate college (Odds Ratio 2.23, 95% Confidence Interval 1.46–3.41) and had more moderate to severe neurologic conditions (18.4% v 8.2%, p<0.001), which were associated with poor memory (β=−0.90, p<0.001), attention (β=−1.02, p<0.001), and HRQOL general health (β=−.80, p=0.0015). Treatment variables and cardiopulmonary morbidity (higher in survivors) were not associated with outcomes.

Conclusions:

Survivors of WT demonstrate impairment in neurocognitive function and have lower social attainment during adulthood, with poorer neurocognitive function associated with neurologic morbidity.

Introduction

Survival rates for childhood cancer have increased dramatically during the last 50 years, with 83% currently surviving at least 5 years [1]. The majority of long-term survivors experience late sequelae, including physical, neurocognitive and/or psychosocial morbidity [2–4]. Neurocognitive sequelae have mostly been studied in survivors of brain tumors and acute lymphoblastic leukemia [5–7], for whom CNS-directed therapies are used (e.g. cranial radiation therapy and/or intrathecal/high dose intravenous methotrexate). However, recent literature demonstrates long-term survivors also experience neurocognitive problems due to chronic health conditions that develop in response to cancer therapy (e.g. cardiopulmonary morbidity) [8, 9]. Also, there is emerging evidence that non-CNS-directed chemotherapy might affect the brain through endocrine changes or immune dysregulation [10].

Wilms tumor (WT) is a kidney tumor with an annual incidence rate of 9 per million population under the age of 15 years [1]. Median age of diagnosis is 3–4 years and five-year survival rates are 91%. Genetic anomalies and syndromes are relatively common in WT patients, affecting 9–17% [11]. Historically, treatment for WT included surgery and chemotherapy in most cases together with abdominal radiation in more than half of the patients [12]. By 50 years of age, long-term survivors of WT have, on average, 14 chronic health conditions, including four that are serious/disabling, life-threatening or fatal [12]. In two reports from the Childhood Cancer Survivor Study, Termuhlen et al. [13] found 65% of WT survivors to have a least one chronic health condition, including cardiopulmonary and renal problems, and Mostoufi-Moab et al. [14] reported 29% to have endocrinopathies. However, very few studies have examined neurocognitive outcomes in long-term WT survivors, with those available studies including very small samples with short follow-up (less than 10 years post diagnosis) [15–18]. Two studies with longer follow-up time (10 and 21 years post diagnosis respectively) found lower neurocognitive abilities; Mulhern et al. [19] reported that a group (n=26) of WT survivors scored 0.5–1.0 standard deviations below population means on verbal memory tests and Mackie et al. [20] found survivors of WT to have poorer functioning in social skills and coping ability compared to healthy controls. Assessment of academic outcomes among WT survivors have provided inconsistent results [21–25]. Poor emotional and physical health in survivors of WT have also been reported, especially in survivors receiving pelvic radiation [13, 26], which might cause poorer neurocognitive long-term outcome since cardiopulmonary, renal and endocrine deficits have been shown to be associated with cognitive impairment in non-cancer patients [27–29].

To address the gaps in knowledge regarding neurocognitive outcomes among long-term survivors of WT, and to investigate potential association with cardiopulmonary, renal and/or endocrine morbidity, we undertook direct neurocognitive testing in a large cohort of well characterized adult survivors of WT. The aims of this study were to evaluate neurocognitive function, health-related quality of life (HRQOL), emotional distress, and social attainment in adult survivors of WT and to identify demographic and treatment-related predictors of outcomes.

Methods

Participants

This current analysis utilizes participants in the St Jude Lifetime Cohort study (SJLIFE), a retrospective cohort with prospective longitudinal clinical assessment of five-year survivors treated for childhood cancer at St Jude Children’s Research Hospital [30, 31]. Survivors are followed prospectively with medical, psychosocial and neurocognitive evaluations. Neurocognitive testing was risk-based when the study started in 2007, but since April 2012 all survivors were recruited for neurocognitive testing, i.e. survivors visiting campus before April 2012 were considered non-eligible for the present study (n=48). Survivors more than 10 years from diagnosis were included in the current analysis. A total of 236 WT survivors were eligible for the present study of whom 158 (67%) had been assessed for neurocognitive outcomes (Fig 1). A control group of 354 individuals without a history of childhood cancer, recruited from the community and from friends and non-first-degree relatives of St Jude patients, was included. Exclusion criteria for both the survivors and the control group were pre-existing non-cancer-related genetic or neurodevelopmental disorder associated with neurocognitive impairment, brain injury unrelated to cancer, non-proficiency in English, low birth weight <750g, and CNS involvement (WT survivors). Seven survivors (four with pre-existing conditions, one with severe prematurity and two with CNS involvement) and 14 controls were considered non-eligible due to those criteria.

Figure 1.

CONSORT diagram survivors

Measures

The neurocognitive tests covered six domains: intellectual function, academics, attention, processing speed, memory and executive function (Supplemental Table S1). Analyses included subtests from the Wechsler Abbreviated Scale of Intelligence [32], Woodcock Johnson Tests of Achievement III NU [33], Conners’ Continuous Performance Test II [34], California Verbal Learning Test – 2^nd edition [35], Trail Making Test [36], Test of Memory and Learning II) [37], and Controlled Oral Word Association Test [38]. Testing was conducted by certified neuropsychological examiners under the supervision of a board-certified neuropsychologist, according to a structured protocol designed to reduce distractions and fatigue.

Assessment of HRQOL and emotional distress was obtained using the 36-item Short Form survey (SF-36) [39], and the Brief Symptom Inventory 18 (BSI-18) [40], respectively. The SF-36 includes eight indices (physical functioning, role limitations due to physical problems, bodily pain, general health, vitality, social functioning, role limitation due to emotional problems, and mental health), which combine into two global scales (physical component summary and mental component summary). The BSI-18 provides subscales for anxiety, depression, and somatization as well as an overall global severity index.

Classification and severity grading of chronic conditions utilized medical records, health questionnaires, physical examinations, and laboratory-based clinical tests [31]. The following social attainment measures were obtained using questionnaires: education (college graduate or higher vs <college graduate), employment (full time employment/student/caring for home vs no or part-time employment), marriage (never vs ever married), household income (<$40 000 vs ≥$40 000), independent living (yes/no), driver’s license (yes/no) and health insurance (yes/no).

Statistical analysis

Measures of neurocognitive ability, HRQOL and emotional distress were converted to z-scores using standardized national norms. For neurocognitive function and HRQOL higher scores indicate better outcome, while for emotional distress, higher scores indicate worse outcome. Demographics of the survivor and the control group were compared using Student’s t-test and χ²-tests, and any significant differences were included as covariates in subsequent analyses between groups. Two-sample T-tests, adjusted for race/ethnicity, were used to compare differences between the study group and the control group for neurocognitive, HRQOL, social attainment and emotional distress outcomes and p-values were adjusted for false discovery rate [41]. To identify clinically relevant impairment, survivors were also compared to population norms using one-sample t-tests. If survivors yielded significantly lower scores compared to both norms and controls on any outcome variable, this variable was included in multivariable generalized linear models. Separate models were built for each outcome, adjusting for sex, race and time since diagnosis.

Results

Characteristics of the study population, including demographic, treatment and chronic health conditions present at the time of neurocognitive assessment are provided in Table 1. A drop-out analysis found no significant differences between participating and non-participating survivors regarding diagnosis, treatment, sex or ethnicity (Supplemental Table S2). Non-participants were slightly older at diagnosis (3.6 years vs 4.6 years, p=0.04). Significant differences between survivors and controls were found for race/ethnicity, with a higher percentage of non-Hispanic whites in the control group. Survivors were diagnosed 1963–2005, with 29 survivors (18.4%) being diagnosed 1975 or earlier. Average age at WT diagnosis was 3.6 years and the majority of survivors underwent unilateral nephrectomy and treatment with vincristine, dactinomycin, and doxorubicin. 68% of survivors received radiation therapy either to the abdomen only (49%) or abdomen and chest (19%). Survivors had significantly more chronic conditions than controls within every organ system, most commonly cardiovascular (40%) and respiratory (39%) conditions. Sixty-one percent had an endocrine condition, mostly overweight/obesity (52%) with only 9% of all survivors had an endocrine condition other than overweight/obesity.

Table 1.

Demographic and treatment characteristics, chronic conditions at assessment

	Survivors N=158	Controls N=354
	M (SD) / N(%)	M (SD) / N(%)	P
Age at evaluation, years	33 (9.1)	35 (10.2)	0.057
Age at diagnosis, years	3.6 (2.6)
Time since diagnosis, years	29 (9.1)
Sex			0.44
Male	64 (40.5)	157 (44.4)
Female	94 (59.5)	197 (55.6)
Race/Ethnicity			<.001
White, non-Hispanic	116 (73.4)	317 (89.6)
Other	42 (26.6)	37 (10.4)
Radiation
No radiation treatment	51 (32.3)
Radiation to the abdomen only	77 (48.7)
Dose when >0 abdomen, cGy	1960 (1010)
Dose when >0 chest (scatter), cGy	270 (240)
Radiation to the abdomen and chest	30 (19.0)
Dose when >0 abdomen, cGy	1740 (1050)
Dose when >0 chest, cGy	1870 (1820)
Other radiation treatment locations	0 (0.0)
Chemotherapy
Vincristine	155 (98.1)
Dose when >0, mg/m2	32.4 (14.9)
Dactinomycin	152 (96.2)
Dose when >0, mg/m2	7.5 (3.6)
Doxorubicin	108 (68.4)
Dose when >0, mg/m2	195.6 (66.6)
Cyclophosphamide	7 (4.4)
Etoposide	21 (13.3)
Carboplatin	14 (8.9)
Other chemotherapy	19 (12.0)
Surgery
Unilateral nephrectomy	147 (93.0)
Partial nephrectomy	1 (0.6)
Bilateral nephrectomy	2 (1.3)
Other	8 (5.1)
Chronic conditions, grade 2–4
Cardiovascular	64 (40.5)	69 (19.5)	<.001
Respiratory	62 (39.2)	39 (11.02)	<.001
Reproductive	48 (30.)	36 (10.2)	<.001
Endocrine (total)	97 (61.4)	169 (47.7)	0.0045
Endocrine other than overweight/obesity	20 (12.7)	10 (2.8)	0.002
Renal	18 (11.4)	14 (4.0)	<.001
Musculoskeletal	51 (32.3)	19 (5.4)	<.001
Neurology	29 (18.4)	29 (8.2)	<.001

Comparison chronic conditions adjusted for age at assessment and race.

WT survivors had significantly lowered scores, relative to normative data and compared to community controls, in verbal reasoning, word reading, mathematics, sustained attention, long-term verbal memory and verbal fluency, with average z-scores ranging from −0.64 to −0.21 (M=0, SD=1) (Table 2). Regarding HRQOL, survivors had lower scores on subfactor general health only (M=−0.22, SD=1.04). Compared to norms, survivors rated themselves having significantly better HRQOL on most physical subfactors, however most ratings were significantly worse than controls. No differences were found regarding mental HRQOL. Survivors reported more emotional distress compared to controls, but reports were not significantly different from norms except that survivors rated themselves having less anxiety. There were no differences between survivors and controls regarding frequency of full-time employment, marriage, independent living, health insurance and driver’s license status. Survivors did have a lower rate of college graduation and lower income, after adjusting for age at assessment and race.

Table 2.

Neurocognitive, emotional and quality of life measures

	Survivors, N= 158			Controls, N= 354
	Mean (SD)	P vs norms		Mean (SD)	P vs controls
Intellectual Function
Verbal reasoning	−0.39 (1.14)	0.004		0.18 (0.90)	0.002
Non-Verbal reasoning	0.013 (1.00)	0.88		0.27 (0.73)	0.045
Overall reasoning	−0.17 (1.02)	0.058		0.27 (0.78)	0.002
Academics
Reading	−0.30 (0.63)	0.004		−0.002 (0.54)	0.002
Mathematics	−0.64 (0.94)	0.004		−0.12 (0.81)	0.002
Attention
Sustained attention	−0.46 (1.37)	0.004		−0.030 (1.10)	0.002
Focused attention	0.19 (1.17)	0.058		0.58 (0.83)	0.002
Processing speed
Visuomotor processing speed	0.046 (1.01)	0.64		0.45 (0.94)	0.002
Cognitive processing speed	0.17 (1.03)	0.076		0.64 (0.91)	0.002
Memory
Short-term memory	−0.10 (0.97)	0.23		0.15 (0.94)	0.063
Verbal learning	−0.015 (1.21)	0.88		0.34 (1.00)	0.045
Long-term verbal Memory	−0.33 (1.31)	0.007		0.10 (1.04)	0.005
Visual memory	−0.19 (1.07)	0.058		−0.065 (0.97)	0.63
Executive function
Working memory	−0.15 (0.81)	0.049		0.019 (0.86)	0.16
Self-monitoring	−0.33 (1.94)	0.058		−0.11 (1.48)	0.20
Cognitive flexibility	−0.28 (1.65)	0.058		0.30 (1.18)	0.002
Verbal fluency	−0.21 (1.09)	0.046		0.079 (1.07)	0.033
Emotional distress BSI
Somatization	0.16 (0.97)	0.10		−0.33 (0.78)	0.004
Depression	−0.10 (1.01)	0.23		−0.34 (0.80)	0.001
Anxiety	−0.22 (1.07)	0.044		−0.41 (0.88)	0.001
Global Severity Index	−0.11 (1.13)	0.23		−0.48 (0.97)	0.001
Health related quality of life SF36
Physical function	0.20 (0.83)	0.013		0.39 (0.70)	0.38
Role limitations due to physical health problems	0.19 (0.91)	0.028		0.44 (0.66)	0.013
Bodily pain	0.29 (1.00)	0.009		0.42 (0.91)	0.20
General health	−0.22 (1.04)	0.028		0.26 (0.91)	0.009
Vitality	−0.13 (1.08)	0.18		0.16 (0.97)	0.022
Social functioning	−0.14 (1.08)	0.18		0.18 (0.85)	0.010
Role limitations due to emotional problems	−0.005 (1.00)	0.96		0.22 (0.79)	0.19
Mental health	−0.12 (1.14)	0.22		0.072 (0.91)	0.050
Physical Component Summary scores	0.21 (0.85)	0.013		0.46 (0.77)	0.38
Mental Component Summary scores	−0.20 (1.12)	0.063		0.030 (0.89)	0.010
	N	%	N	%	P vs controls
Social attainment
College graduate or higher	51	34.23	178	54.94	0.001
Working full/part-time	111	77.08	278	85.28	0.33
Ever married	105	70.95	271	79.71	0.33
Household income >=40,000	36	24.66	148	44.05	0.007
Living independently	93	64.14	260	76.25	0.13
Health insurance	126	85.71	297	87.35	0.88
Driver’s license	141	95.27	328	96.18	0.88

One or two-sample T-test, adjusted for race/ethnicity. Social attainment also adjusted for age at assessment. Bold type indicates variables for which survivors performed significantly lower than both normative data and controls. All measures in z-scores: M=0, SD=1. P-values adjusted for false discovery rate.

Multivariable models adjusted for age at diagnosis, race and time since diagnosis demonstrated few associations between treatment variables and cognitive and HRQOL variables (Table 3). Subsequent exploratory analyses showed an association to era of diagnosis, with patients treated earlier (1960–1975) receiving more radiation and higher doses of chemotherapy and also demonstrating better neurocognitive outcome. Being non-Hispanic and white was associated with higher scores on neurocognitive tests, but race/ethnicity was not associated with social attainment or HRQOL.

Table 3.

Multivariable models for associations between cancer treatment, chronic conditions and cognitive and HRQOL outcome

		Verbal reasoning		Word reading		Mathematics		Sustained attention		Long-term verbal memory		Verbal fluency		HRQOL General health
		Est.	P-val	Est.	P-val	Est.	P-val	Est.	P-val	Est.	P-val	Est.	P-val	Est.	P-val
Ridge regression	Dactinomycin	0.001	0.91	0.010	0.056	0.018	0.086	0.013	0.36	0.072	0.014	0.004	0.77	0.058	0.62
	Vincristine	−0.0001	0.95	0.0003	0.75	−0.001	0.65	−0.001	0.78	−0.010	0.20	0.001	0.81	0.007	0.62
	Doxorubicin	−0.0002	0.54	−0.0003	0.15	−0.001	0.12	−0.0002	0.76	−0.002	0.11	−0.0004	0.49	−0.0003	0.74
	Radiation
	Radiation to abdomen	0.090	0.28	−0.019	0.66	0.14	0.097	−0.13	0.28	0.55	0.0062	0.12	0.30	−0.16	0.45
	Radiation to abdomen and chest	0.021	0.84	0.028	0.62	0.019	0.86	0.20	0.21	0.46	0.070	0.40	0.0073	−0.73	0.004
	Chronic conditions Grade 2–4
Generalized linear models	Cardiovascular system	0.076	0.70	0.13	0.19	0.014	0.93	0.32	0.19	0.012	0.96	−0.10	0.59	−0.34	0.081
	Respiratory system	0.080	0.67	0.054	0.59	0.022	0.89	0.18	0.45	0.23	0.30	0.20	0.28	−0.23	0.22
	Reproductive system	−0.13	0.52	−0.041	0.70	0.123	0.47	0.15	0.56	−0.051	0.83	−0.17	0.39	−0.46	0.021
	Endocrine system	−0.34	0.053	−0.11	0.25	−0.26	0.084	0.15	0.52	−0.23	0.27	−0.15	0.39	−0.44	0.012
	Renal system	0.45	0.11	0.023	0.88	0.25	0.29	−0.23	0.51	0.28	0.40	0.25	0.37	−0.66	0.016
	Musculoskeletal system	0.12	0.54	−0.015	0.88	0.044	0.78	−0.56	0.016	0.083	0.70	−0.061	0.74	−0.21	0.25
	Neurology system	−0.44	0.052	−0.18	0.15	−0.30	0.11	−0.73	0.010	−0.78	0.0025	−0.27	0.22	−0.80	<0.001
	Second neoplasms (Yes)	0.49	0.10	0.24	0.13	0.25	0.33	0.41	0.29	0.22	0.55	0.22	0.46	0.22	0.46

Ridge regression models for treatment variables, generalized linear models for chronic conditions. Separate models for each cognitive outcome variable. Chemotherapy and radiation in the same models, chronic conditions in separate models. Non-standardized estimates. Cognitive measures in z-scores. Chemotherapy per mg/m². Radiation as compared to no radiation. All models adjusted for sex, race/ethnicity and time since diagnosis. Bold type indicates significant models.

Associations were found between the presence of musculoskeletal conditions and poorer sustained attention and between the presence of neurologic conditions and poorer sustained attention and long-term verbal memory (Table 3). Reduced HRQOL, specifically general health, was associated with several chronic conditions; reproductive, endocrine, and renal as well as neurologic conditions. No associations were found between chronic conditions and social attainment. Further analyses of neurologic conditions showed survivors have a higher prevalence of motor (3.8% vs 0%, p<0.001) and sensory neuropathy (22.8% vs 9.0%, p<0.001) compared to controls (Supplemental Table S3).

Higher scores on measures of verbal reasoning, word reading, and mathematics were associated with higher social attainment and higher scores on measures of verbal memory alone were associated with better HRQOL (Table 4).

Table 4.

Cognitive outcome associated to HRQOL and social attainment

	Education College graduate or higher		Household income		SF-36 Social functioning		SF-36 Mental health
	OR	P-val	OR	P-val	Est.	P	Est.	P
Verbal reasoning	2.23	<0.001	1.59	0.029	0.11	0.17	0.12	0.15
Word reading	3.47	0.0025	2.82	0.024	0.11	0.47	0.25	0.12
Mathematics	2.55	<0.001	1.74	0.050	0.098	0.33	0.17	0.11
Sustained attention, CPT variability	1.59	0.0067	1.12	0.51	0.11	0.12	0.10	0.14
Long-term verbal memory CVLT LD	1.35	0.045	1.24	0.23	0.21	0.002	0.27	<0.001
Initiation: verbal fluency FAS	1.289	0.14	1.04	0.86	0.081	0.32	0.13	0.14

Separate models for each cognitive variable and each outcome variable. Generalized linear models for HRQOL outcomes, logistic regression models for social attainment; reference category least favorable outcome (<college graduate, household income <$40000/year). Bold type indicates p<0.05

Discussion

Long-term survivors of WT demonstrate lower neurocognitive function, including verbal reasoning, academics, attention, memory and executive function. Lower performance in these neurocognitive functions is associated with lower social attainment (income and education) and lower social and mental HRQOL. Poor functional outcomes were associated with the presence of current chronic neurological and musculoskeletal conditions, with survivors demonstrating higher frequency of these chronic conditions compared to controls. Survivors did not show elevated levels of emotional distress or mental HRQOL.

This is the largest study of adult survivors of WT using objective neurocognitive testing. Several previous studies have used smaller samples of survivors of Wilms tumors as a control group when studying cognition in survivors treated for acute lymphoblastic leukemia [18, 19]. Most of those studies did not find differences between survivors of WT and controls. However, it is unclear whether a representative sample of WT survivors was included, as the primary emphasis was on leukemia survivors. When impairment was found in previous studies, the findings have been consistent with our results showing problems in memory and academics [19, 20, 25].

Neurocognitive problems in survivors included in the current study were mainly within verbal skills (verbal reasoning, word reading, verbal memory and verbal fluency). Survivors of childhood cancer treated with cranial radiation or intrathecal chemotherapy often demonstrate attention, memory, processing speed and executive function problems, but not purely within the verbal domain [42–44]. Survivors’ lower performance in verbal skills might be linked to the consistently young age at diagnosis (median=3.2 years, interquartile range 1.8–4.7 years), an age known for intense verbal development when any negative impact on development could affect those skills. A similar pattern and profile were found in children with early onset diabetes [45, 46].

Survivors had a higher prevalence of chronic conditions, including cardiovascular, respiratory and endocrine conditions, compared to community controls. Contrary to expectation none of these conditions were related to worse neurocognitive function in multivariable models. Neurologic chronic conditions were also more common in survivors than in controls (18% vs 8%), mainly a higher prevalence of peripheral motor and sensory neuropathy, a common side effect to treatment with vincristine [47]. Having a neurologic condition was closely linked to lower scores on neurocognitive tests and HRQOL measures (none of which were strongly motor-based). Peripheral neuropathy is generally considered an adverse effect to chemotherapy, but several recent studies have pointed out that it might be both long-lasting and associated with severe disability and reduction of quality of life [48]. Peripheral neuropathy might, for example impair a person’s ability to dress, to drive a car or it could impair ability to take part in social activities [49, 50]. In individuals suffering from diabetes, peripheral neuropathy has been linked to depression, mediated by psychosocial consequences [51, 52]. Clearly peripheral neuropathy is a late effect that needs more attention and support in long-term survivors. An association between long-term peripheral neuropathy and development of neurocognitive function has not been shown before.

In rare cases, Vincristine-induced peripheral neuropathy is also associated with central and autonomic nervous system dysfunction, leading to more global neurological symptoms; e.g. headache, confusion, seizures, or visual disturbances, sometimes together with transient neuroimaging findings [53–56]. Such comorbid neuropathy has also been associated with other conditions and treatment factors such as hypertension, treatment with immunosuppressants, or acute renal failure [53, 57]. Even if such central and autonomic neuropathy is rare a milder version may account for the association between peripheral neuropathy and cognitive impairment in this study, either as a side effect to treatment or following later renal or hepatic chronic conditions. Future studies should prospectively examine peripheral autonomic and central neuropathy and cognitive development in patients treated for Wilms tumor.

Few associations were found between outcomes and treatment variables. For the most part survivors received very similar treatment at a similar diagnostic age, and this lack of variability might be one reason for the lack of associations. We did not expect to see an association between higher doses of chemotherapy and abdominal radiation and better neurocognitive outcomes. This association remained significant even after sensitivity analysis. Higher chemotherapy combined with abdominal radiation was more common in survivors who were treated in the 1960’s and 1970’s compared to the 1980’s and 1990’s. The association might therefore represent a survival effect, i.e. improved supportive care on more recent protocols may have resulted in sicker children living to long-term survival compared to only relatively healthy children surviving on earlier protocols.

The results of this study should be interpreted in light of limitations. First, all survivors were treated during a long period of time and at a single institution and the results might have limited generalizability to survivors treated more recently and at other institutions. Still, this is the largest study directly assessing long-term survivors of WT and protocols employed are similar to those employed at other institutions. Also, those long-term survivors exist and need support. Prospective studies, including pre-therapy testing, might permit identification of trajectories of growth and development in survivors treated on different protocols. Second, the control group, including fewer minorities and slightly more well performing individuals, might not be comparable to a population-based sample, leading to an over-estimation of impairment found among survivors. The higher percentage of black survivors is in line with previous studies finding the prevalence of Wilms tumor to be slightly higher among black children (29% black children diagnosed with Wilms tumor vs 18.5% in other childhood cancers) [58], and all multivariate models were adjusted for race. Also, a stricter criterion for impairment (performance worse than both norms and controls) was applied. Both survivors and controls were slightly younger than a population-based sample, which might be a reason for elevated HRQOL and lower emotional distress.

Despite these limitations, our results demonstrate that long-term survivors of WT show neurocognitive impairment in verbal domains that impact social attainment and health-related quality of life. Survivors who have neurologic chronic conditions seems especially vulnerable and more prospective research is needed to study the effects of peripheral neuropathy on cognition and HRQOL. Neurocognitive and social outcomes need to be assessed earlier in survivorship and appropriate rehabilitation offered prior to adult outcomes. Since this study is one of the first of its kind, further prospective research is needed to confirm these results and characterize the developmental trajectory by identifying the onset and progression of the impairment over time.

Implications for Cancer Survivors:

Survivors of WT should be offered neurocognitive evaluations and rehabilitation. Neurologic conditions should be routinely assessed, and appropriate support offered to reduce risk for functional limitations.